RADIO SCIENCE, VOL. 41, RS5S31, doi:10.

1029/2005RS003356, 2006

Perils of the GPS phase scintillation index (SF)

Theodore L. Beach1
Received 30 August 2005; revised 26 January 2006; accepted 14 February 2006; published 23 May 2006.
[1] The phase scintillation index (sf), equal to the standard deviation of measured phase,
is often used to characterize Global Positioning System (GPS) observations in
ionospheric environments that may be scintillated. Since sf is dominated by large-scale
fluctuations, questions of usage and interpretation exist as will be illustrated here.
Beyond traditional concerns with detrending, multipath and receiver phase noise, there
are at least two issues to be considered. The first is the marginal suitability of sf to
characterize a power law phase screen with a poorly defined low-frequency component
(e.g., outer scale). Second, observed sf parameters may not be relevant to GPS
receiver tracking impacts. These arguments are outlined here in greater detail and are
illustrated with simple one-dimensional phase screen propagation modeling results. The
conclusion is that GPS sf values depend critically on the circumstances of measurement
and are difficult to compare among observations without additional knowledge,
particularly of relative ionospheric drift and irregularity orientation, that may not be
available from an isolated GPS receiver. The development of suitable alternative
measures requires careful consideration of the elements of GPS scintillation and its
impacts. The broader GPS scintillation community should take an active role in
developing suitable replacement measures for sf.
Citation: Beach, T. L. (2006), Perils of the GPS phase scintillation index (sf), Radio Sci., 41, RS5S31,
doi:10.1029/2005RS003356.

1. Introduction reappraisal of the continued use of sf as a quantitative

measure, particularly in GPS studies.
[2] The phase scintillation index (sf), defined as the [3] That phase fluctuations have fundamental impor-
standard deviation of measured phase, has historically tance to ionospheric scintillation is not in dispute.
been used to characterize the phase component of iono- Indeed, perturbations in phase at the ionosphere are the
spheric scintillation [Fremouw et al., 1978; Yeh and Liu, source of the developed diffraction pattern in the phase
1982]. With the advent of Global Positioning System screen model. At issue, rather, is the utility of character-
(GPS) scintillation monitoring, the research community izing ionospheric phase by a single parameter that masks
has tended to apply traditional measures like sf to crucial information such as fluctuation power at partic-
the newer observations. Previous analysis [Forte and ular scale sizes. The discussion of utility will be
Radicella, 2002] demonstrated some of the inadequacies informed by several perspectives: (1) applicability to
of using the traditional phase scintillation index with propagation studies, (2) observational limitations and
GPS. A key problem is unknown relative motion (3) relevance to receiver impacts. In order to keep the
between the ionosphere and the GPS line of sight arguments focused on the most fundamental concepts
(LOS) coupled with the fact that the bulk of phase they will be illustrated with one-dimensional (1-D) phase
fluctuation power due to ionospheric irregularities screen examples, although the discussion will mention
resides at the larger spatial scales. The goal of this paper two-dimensional (2-D) effects where appropriate. The
is to expand on some of those concerns and to suggest upshot will be that suitable phase fluctuation measures
must account for spatial scales and velocity effects
pertinent to the reasons chosen to quantify phase fluctu-
1
Space Vehicles Directorate, Air Force Research Laboratory,
ations in the first place.
Hanscom Air Force Base, Massachusetts, USA. [4] Many, but not all, of these considerations stem
from the fact that the magnitude of relative ionospheric
This paper is not subject to U.S. copyright. motion of the GPS LOS typically falls into an interme-
Published in 2006 by the American Geophysical Union. diate regime compared to other satellite beacon measure-
RS5S31 1 of 7
RS5S31 BEACH: GPS PHASE SCINTILLATION INDEX PROBLEMS RS5S31

Figure 1. Phase screen geometry.

ments. It is not possible to assume with GPS that the to theoretical studies has been that of the phase at the
ionosphere is essentially fixed during the satellite pass as phase screen [Salpeter, 1967], sometimes denoted f20 or
is the case with a low Earth orbit (LEO) beacon; neither by other notation. In practice, the variance of measured
is it possible to assume that the satellite link is fixed and phase at the ground is used as a proxy since the
all motion is due to the ionosphere as with a geostation- equivalent phase screen representing the ionospheric
ary beacon. There may be cases of observations from state is not available. This paper will adopt the
moving receivers where some of these simplifying convention that sf refers to the phase at the screen,
assumptions could apply, for example GPS measure- while the standard deviation of the phase of the
ments from LEO satellites, but in general they do not. propagated wavefront is sfp.
[5] Finally, note that some of the observational factors [7] Figure 1 illustrates the geometry for a 1-D phase
to be discussed may apply to the amplitude scintillation screen and the numerical difference between sf and sfp
index (S4), the root-mean-square deviation of normalized for a particular phase screen realization, an error of about
signal power, as well as sf. Fortunately, spatial-scale 20% in this case. The expression for 1-D phase screen
structures much larger than the Fresnel radius do not propagation is
influence S4 as strongly as they do sf unless a significant "   #
Z1
refractive component to the scintillation exists. Discus- ifp ð xÞ 1 0 ifð xÞ x  x0 2
sion of observational and operational concerns regarding Ap ð xÞe ¼ dx e exp ip ; ð2Þ
rF rF
S4 will be considered beyond the scope of the present 1
text. iwt
where e time dependence is implied and constant
phase factors in front of the integral have been omitted.
2. Definitions The quantities A p (x) and sp (x) are, respectively,
[6] In mathematical form, the phase scintillation index the amplitude
pffiffiffiffiffi and phase of the propagated wave while
is rF = lz is the Fresnel radius, where l is the radio


wavelength and z is the perpendicular distance away
s2f ¼ f2  hfi2 ; ð1Þ from the phase screen that the wavefront has propagated.
Note that the phase sp(x) has been ‘‘unwrapped’’ for
where sf is phase and the angled brackets h. . .i denote purposes of presentation and of computing sfp. Also,
expected value, usually replaced by temporal averaging spatial averaging is used to estimate the expected values
in practice. Already this definition incorporates some in equation (1). Inspection of Figure 1 clearly shows that
deliberate haziness to make a point; namely, which phase larger-scale structures dominate in their contribution to
is to be used? In principle, the phase variance of interest sf and sfp for this realization, a power law phase screen

2 of 7
RS5S31 BEACH: GPS PHASE SCINTILLATION INDEX PROBLEMS RS5S31

Figure 2. Power law (top) spectra and (bottom) phase realizations.

example with ‘0 = 8.0 km to be detailed in the next scale, ‘0. These phase screens began with identical
section on propagation studies. A further upcoming realizations of Gaussian-distributed, band-limited white
section will discuss effects related to temporal averaging noise at 0.19-m sample spacing (=l1, the wavelength of
of the received phase and its velocity dependence. the GPS L1 carrier) that were then spectrally shaped. The
mathematical form of the power spectral density (PSD)
used is
3. Propagation Studies 8 q
> C; < ‘1
[8] Rino [1982, p. 860] notes that ‘‘most theoreticians q
> < 2p 0
have used’’ a parameter equivalent to sf as a ‘‘strength F ¼  p ð3Þ
parameter’’ but that this usage ‘‘can be misleading’’ 2p >
> q‘0 q
:C ;
‘1
0 ;
because it is possible to construct screens with different 2p 2p
power spectra and a broad range of sp values that give
essentially identical signal strength patterns. Rather than where the exponent p is the spectral index, q is the spatial
delve into the theoretical details of his brief argument we frequency expressed in radians per meter and C is a scale
shall examine the effect through concrete examples. factor. The values of the spectral index and outer scale
[9] Consider two realizations of 1-D power law phase are listed in the PSD plots (top plots). Note that, as
screens depicted in Figure 2, where f(x) is the phase at intended, the spectra match above the q corresponding to
the screen as in equation (2) above. The comparison ‘0, with the exception of a slight tendency toward
visually demonstrates the qualitative differences between developing a noise floor in the ‘0 = 8 km case. In fact,
power law realizations having different outer scales; this tendency, coupled with the 50-km total length of the
specifically, the bulk of the fluctuation power appears screen, precludes the use of an outer scale much larger
at larger scale sizes for the realization with a greater outer than about 10 km here. The total number of points in the

3 of 7
RS5S31 BEACH: GPS PHASE SCINTILLATION INDEX PROBLEMS RS5S31

Figure 3. Comparative received power plots (enlargement, bottom curve offset by 20 dB).

screen, 218, is a compromise between computer memory GPS phase is often derived from dual-frequency total
constraints and allowing sufficient length to vary ‘0, electron content (TEC) estimates, which contain errors
given the fundamental 0.19-m point spacing. due to the frequency dependence of phase scintillation
[10] Figure 3 compares the results for signal power, [e.g., Bhattacharyya et al., 2000].
P(x) = jAp(x)j2, for the two phase-screen realizations
obtained from FFT-based computations [Buckley, 1975]
of the 1-D propagation integral (2). The power plots are
4. Temporal Sampling
qualitatively very similar; many common features be- [13] Greater concern develops with the conversion of
tween the two can be identified. Quantitatively, in both spatial diffraction patterns into the time series that GPS
cases S4 = 0.800. Other calculation details include the receivers actually record. A recent series of papers
assumption of a zenith phase screen atpaffiffiffiffiffiffiffi
distance z = [Kintner et al., 2001, 2004; Ledvina et al., 2004]
300 km, giving a Fresnel scale rF = l1 z = 240 m. describes some of these factors for the equatorial region
Thus the outer scales are greater than the Fresnel scale and illustrates them with observational examples. Factors
for both phase screens presented. for a stationary receiver include LOS geometry relative
[11] The immediate conclusion is that sf, which is to the magnetic field, ionospheric drift, satellite motion
dominated by the larger scale variations in phase, does and the temporal evolution of the irregularities. At
not bear close relation to amplitude scintillation levels if auroral or polar cap latitudes the situation becomes even
the phase variance in a band about the Fresnel wave more complicated because of varying 2-D configurations
number is relatively constant. It is possible to tailor sf of ionospheric structure [e.g., Livingston et al., 1982].
almost at will for a given S4 by altering the low- This discussion will consider only the effects of net
frequency portion of the PSD, to include spectral breaks ionospheric drift speed, and then in a rather simplistic
as well as a relatively arbitrary outer scale. Beach and fashion, to make some illustrative points.
Kintner [1999] model and discuss the sensitivity of the [14] Forte and Radicella [2002] treat the effects of a
dependence of the relationship of S4 to sf on ‘0 for a fixed (temporal) frequency filter cutoff in the phase
PSD similar to Equation (3) in the weak scatter approx- detrending process for variable relative drift speeds.
imation. (Jokipii [1970] proves that the weak scatter They demonstrate in a semiquantitative fashion that the
approximation is valid up to higher S4 values than is drift, usually unknown for a single-receiver setup, can
perhaps commonly appreciated; one requirement is that have a significant impact on temporally averaged sfp.
the PSD fall off sufficiently rapidly.) Conceptually, this result should be quite evident.
[12] Again, what is measured as ‘‘sf’’ by a GPS Changes in the relative drift cause the phase fluctuation
receiver is in fact related to sfp, not sp. This approxi- spectrum to shift left or right. In the mean time, the
mation is perhaps not too far wrong in the examples detrending process will remove a portion of the low-
above. One can compute the standard deviation of frequency power from the fluctuation spectrum; the
propagated field phase directly, after unwrapping, to find power removed depends critically on where the detrend-
sfp = 12.5 rad in the ‘0 = 8.0 km example (cf. Figure 1) ing cutoff falls relative to the shifted spectrum. Since
and sfp = 4.5 rad, an error of a factor of 3, in the ‘0 = low-frequency power dominates phase’s standard devia-
0.8 km case. Another observational consideration is that tion in a power law environment, observed sfp will vary

4 of 7
RS5S31 BEACH: GPS PHASE SCINTILLATION INDEX PROBLEMS RS5S31

Figure 4. Effects of relative drift on the temporally averaged phase scintillation index.

even if the phase fluctuation spectrum remains constant. irregularities [Kil et al., 2000] or ‘‘fade stretching’’ due
If the low-frequency phase spectrum varies under differ- to satellite motion [Kintner et al., 2001]. It should be
ent ionospheric conditions (due to, for example, storm noted that perfectly ‘‘frozen in’’ irregularities have been
time effects [Forte and Radicella, 2002]), additional assumed and that the zenith geometry is particularly
variability in measured sfp will be introduced even as simple. In a realistic case, with irregularities evolving
the amplitude fluctuations remain substantially the same, in time as they drift past, an unknown outer scale
as in the above demonstration. (indeed, an unknown lower-frequency portion of the
[15] An even more basic effect not considered by Forte irregularity spectrum) and LOS motion relative to com-
and Radicella [2002] is implicit high-pass filtering due plex and variable multidimensional structures, the quan-
to the period over which sfp is computed. Figure 4 titative meaning of temporally averaged sfp becomes
shows sequences of temporally averaged sfp estimates quite obscured. Thus the use of temporally averaged
obtained from the received phase of the ‘0 = 8 km case GPS sfp by itself provides a highly inadequate charac-
above using two different assumed relative drift speeds, terization of the irregular ionosphere.
vdrift, and a fixed averaging period of 1 min. Here we
treat the LOS as fixed and consider the ionosphere to be
moving horizontally with the speed vdrift. For simplicity,
5. Receiver Performance
the irregularities do not evolve with time; that is, we [17] The discussion of the implicit filtering in sfp due
assume ‘‘frozen in’’ irregularities. In the left-hand case of to a fixed measurement period above leads naturally to a
Figure 4, the standard deviation of phase is computed discussion of receiver impacts. Intuitively, it does not
directly over the simulated 1-min data segments (i.e., seem likely that the portion of phase variance due to
number of data points = nearest integer to vdrift  60 s/Dx, fluctuations at frequencies lower than a few tenths of a
where Dx is the grid spacing, 0.19 m). Recall that the Hz (i.e., periods of tens of seconds) will have much
value computed above by spatially averaging over influence on the ability of a typical GPS receiver to track.
the entire propagated wavefront is sfp = 12.5 rad. In Nevertheless, the sfp estimates over 1-min periods
the right-hand case, a linear trend has been subtracted include a significant amount of fluctuation power at
from each 1-min data segment before computing sfp these longer timescales; if it were available, the spatially
for that segment. Note that the 50-km-wide phase averaged sfp would contain even more power for the
screen contains fewer total minutes of data as the reasons given above. Careful reading of the work of
relative drift increases. Conker et al. [2003] shows that they incorporate a
[16] The upshot is, the slower the relative drift of this tracking-loop transfer function with a roll-off at low
simple 1-D ionosphere, the lower the temporally aver- frequencies as a weighting factor in the integral of the
aged sfp estimate and linear detrending reduces the value phase PSD that forms their phase scintillation term, s2fs.
further. The range of drift speeds used above is consistent Unfortunately, they call this term the ‘‘phase scintilla-
with typical equatorial cases and that range can be tion’’. A casual reading might then give the false
extended further in many cases, for example, postmid- impression that the variance thus symbolized is the
night zonal drift slowdown or reversal for equatorial square of the usual phase scintillation index. It should

5 of 7
RS5S31 BEACH: GPS PHASE SCINTILLATION INDEX PROBLEMS RS5S31

Figure 5. Effect of receiver motion against a stationary ionospheric scintillation pattern.

be strongly emphasized that any phase variance measure ences, there may not even be wide awareness of them
of scintillation used in receiver tracking analysis must be in the ionospheric GPS community.
appropriately frequency weighted.
[18] Another way to examine this effect is to consider a
hypothetically stationary GPS satellite and a constant and
6. Conclusions
motionless 1-D ionosphere, in this case defined by the [20] The GPS phase scintillation index has many
‘0 = 8.0 km example above, with receivers that are problems as a quantitative measure of ionospheric phase
moving relative to it. Figure 5 depicts the received phase fluctuation—primarily due to the dominance of the low-
(fp) over 10-s segments at two different receiver speeds. frequency component of the phase power spectrum in
Clearly, the 300-m/s (583-knot) case appears that it could its makeup. First, sf can be nearly independent of the
potentially be more stressful to receiver tracking, partic- level of amplitude scintillation as it appears on the
ularly around the 9-s mark, than the 75-m/s (146-knot) ground. Furthermore, the same ionosphere can produce
case. Yet the spatially averaged sf (at the ionosphere) or different observed phase variance levels under different
sfp (at the receiver altitude) is identical between the two relative drift conditions due to implicit high-pass filter-
cases. Not only is there potential for ‘‘fade stretching’’ at ing. Second, phase variance must be appropriately
low relative drifts to degrade receiver performance frequency weighted to parameterize receiver impact.
but also ‘‘phase compression’’ at high relative motion. When computed under ideal observing circumstances,
Critically, the phase scintillation index alone cannot be the usual phase scintillation index incorporates no
used to characterize this effect; characterization requires frequency weighting. This presentation has illustrated
detailed knowledge of the spatial phase fluctuation each of these problems with concrete, if simplistic,
spectrum and relative motion. examples and adding realistic complications does not
[19] An extreme example of such ‘‘phase compres- improve the prognosis. Finally, GPS sf has many
sion’’ may account for observations of space shuttle GPS additional observational concerns not treated here: mul-
velocity errors in the equatorial regions [Kramer and tipath, receiver phase noise and oscillator stability, cycle
Goodman, 2001; Goodman, 2002]. Here the GPS slips under scintillation conditions, large phase trends
receiver moves at about 7 km/s relative to the iono- due to Doppler shifts, use of TEC as a proxy for phase,
sphere. The shuttle typically orbits at bottomside alti- etc. For these reasons it is recommended that the use of
tudes where amplitude scintillation will not yet be sf be seriously reconsidered by the GPS scintillation
developed in response to phase perturbations from an community.
irregular ionosphere overhead. Nevertheless, the large [21] Nevertheless, GPS phase, or related TEC, fluctua-
relative motion undoubtedly leads to additional phase tions should not be overlooked since they can indicate
tracking stress as discussed above. It is conceivable that physical structuring in the ionosphere even in the
this effect may even cause tracking loop stress in cases absence of significant GPS amplitude scintillation. The
where the associated phase perturbations might be question is one of a suitable measure. Alternative mea-
regarded as inconsequential in stationary ground obser- sures to sf must account for the spectral characteristics
vations. Unfortunately, no correlative studies have been of the phase fluctuation as appropriate to its intended use.
performed and, since the observations have not appeared For relating phase fluctuations to amplitude scintillation,
in the ionospheric literature or at geophysics confer- a parameter like the rate-of-TEC index (ROTI) defined
6 of 7
RS5S31 BEACH: GPS PHASE SCINTILLATION INDEX PROBLEMS RS5S31

by Pi et al. [1997] has improved stability under typical Forte, B., and S. M. Radicella (2002), Problems in data treat-
equatorial conditions [Beach and Kintner, 1999], al- ment for ionospheric scintillation measurements, Radio Sci.,
though there has been some inconsistency in defining 37(6), 1096, doi:10.1029/2001RS002508.
the ROTI averaging interval (A. Coster, private commu- Fremouw, E. J., et al. (1978), Early results from the DNA wide-
nication, 2005). A recently suggested alternative may be band satellite experiment—Complex signal scintillation,
to store subsets of phase samples and later use them to Radio Sci., 13, 167.
estimate S4 from a ‘‘sub-sampled phase screen’’ [Beach Goodman, J. L. (2002), Parallel processing: GPS augments
et al., 2004], although the practical details of this TACAN in the space shuttle, GPS World, 13(10), 20.
technique have not yet been fully explored. For receiver- Jokipii, J. R. (1970), On the ‘‘thin screen’’ model of interpla-
tracking impacts, the frequency-weighted phase variance netary scintillations, Astrophys. J., 161, 1147.
of Conker et al. [2003] is a better parameter, although Kil, H., P. M. Kintner, E. R. de Paula, and I. J. Kantor (2000),
it must be recomputed if the relative drift changes and Global Positioning System measurements of the ionospheric
for different receivers with different tracking loop transfer zonal apparent velocity at Cachoeira Paulista in Brazil,
functions. In fact, in some cases the best alternative J. Geophys. Res., 105, 5317.
may be to have representative snapshots of the phase Kintner, P. M., H. Kil, T. L. Beach, and E. R. de Paula (2001),
time series or PSD available together with estimates Fading time scales associated with GPS signals and potential
of the net drift and spatial structuring. Ultimately, consequences, Radio Sci., 36, 731.
the GPS scintillation community should carefully inves- Kintner, P. M., B. M. Ledvina, E. R. de Paula, and I. J. Kantor
tigate, recommend and standardize appropriate phase (2004), Size, shape, orientation, speed, and duration of GPS
fluctuation measures for particular observing situations equatorial anomaly scintillations, Radio Sci., 39, RS2012,
and applications. doi:10.1029/2003RS002878.
Kramer, L., and J. L. Goodman (2001), Global scale observa-
[22] Acknowledgments. The author thanks A. Coster and tions of ionospheric instabilities from GPS in low Earth
P. Doherty for suggesting that the author develop the original orbit, paper presented at AIAA Space 2001 Conference
talk upon which this paper was based. This work was partially and Exposition, Am. Inst. of Aeronaut. and Astronaut.,
supported under AFOSR task 2311AS. Albuquerque, N. M., 28 – 30 Aug.
Ledvina, B. M., P. M. Kintner, and E. R. de Paula (2004),
References Understanding spaced-receiver zonal velocity estimation,
Beach, T. L., and P. M. Kintner (1999), Simultaneous Global J. Geophys. Res., 109, A10306, doi:10.1029/2004JA010489.
Positioning System observations of equatorial scintillations Livingston, R. C., C. L. Rino, J. Owen, and R. T. Tsunoda
and total electron content fluctuations, J. Geophys. Res., (1982), The anisotropy of high-latitude nighttime F region
104, 22,553. irregularities, J. Geophys. Res., 87, 10,519.
Beach, T. L., T. R. Pedersen, M. J. Starks, and S.-Y. Su (2004), Pi, X., A. J. Mannucci, U. J. Lindqwister, and C. M. Ho (1997),
Estimating the amplitude scintillation index from sparsely Monitoring of global ionospheric irregularities using the
sampled phase screen data, Radio Sci., 39, RS5001, worldwide GPS network, Geophys. Res. Lett., 24, 2283.
doi:10.1029/2002RS002792. Rino, C. L. (1982), On the application of phase screen models
Bhattacharyya, A., T. L. Beach, Sa. Basu, and P. M. Kintner to the interpretation of ionospheric scintillation data, Radio
(2000), Nighttime equatorial ionosphere: GPS scintillations Sci., 17, 855.
and differential carrier phase fluctuations, Radio Sci., 35, Salpeter, E. E. (1967), Interplanetary scintillations, I, theory,
209. Astrophys. J., 147, 433.
Buckley, R. (1975), Diffraction by a random phase-changing Yeh, K. C., and C.-H. Liu (1982), Radio wave scintillations in
screen: A numerical experiment, J. Atmos. Terr. Phys., 37, the ionosphere, Proc. IEEE, 70, 324.
1431.
Conker, R. S., M. B. El-Arini, C. J. Hegarty, and T. Hsiao 
(2003), Modeling the effects of ionospheric scintillation on T. L. Beach, Space Vehicles Directorate, Air Force Research
GPS/Satellite-Based Augmentation System availability, Laboratory, Hanscom AFB, MA 01731, USA. (theodore.
Radio Sci., 38(1), 1001, doi:10.1029/2000RS002604. [email protected])

7 of 7